What are agentic workflows, what do they mean for generative AI, and what does it have to do with construction? We begin to explore that this week, looking at how one type of agentic workflow can boost the quality of generative AI content for you. Also, find more bullish reports on AI’s growth in the construction industry, including how applied visual AI is poised to improve safety levels in construction projects.

In this issue, David Dunham introduces one of the newest tools for harnessing generative AI to produce better quality content: Agentic workflows. If you studied social psychology, you might know that “agentic” is a social-psychological phenomenon in which people act as agents or representatives for an authority. But agentic can also mean a mindset of competency. A clue is in the word's root, i.e., "agent." Psychologists talk about a person having "agency," which means the power to act. 

Agentic learning involves self-directed actions aimed at growth and development based on chosen goals. OpenAI defines Agentic AI as systems that can pursue complex goals with limited direct supervision. For generative AI and our purposes here, it is a means to enable AI to produce better content.

Dunham's exploration focuses on one of the four types of agentic workflows: reflection. He walks us through how to use reflection to get better outcomes from generative AI. Along with the exploration, we've included the complete chatlog of his example in PDF form.

Meanwhile, Michael Causey reports on a webinar on how visual AI can improve project safety levels and summarizes the findings of a couple of recent reports from Deloitte and Apiko. It's no big surprise that they are bullish on the application of AI in design and construction.

Good reading.

Regardless of advancements in construction technology, industry leaders are exploring a human-in-the-loop concept that requires the involvement of highly skilled individuals to carry out tasks such as fact-checking, in-depth analysis, and understanding the complex details needed for each construction process. – Deloitte